An Empirical Study of the Function of a Securities Underwriter's Goodwill

Hypothesis and Sampling

Regression Model The following model is designed to check, by regression testing, the functional activeness of an underwriter's goodwill:

Variables: Selection and Definition The function of an underwriter's goodwill is mainly reflected in the accuracy of the valuation of an IPO company, which can be checked with the first-day IPO premium rate. The lower the premium rate, the better the valuation. A low premium rate could indicate that the underwriter is good at disseminating information to investors and its valuation is well-accepted by investors. On the other hand, a high premium rate would mean unsatisfactory fulfillment of the function of goodwill.

In the following formula, IPO represents the first-day premium rate, Px the first-day closing price, and Pt the issue price:

The following five indicators are selected as goodwill variables in this study:

1. Region: Underwriters are divided into nationwide and regional businesses based on the location of their registered offices. An underwriter with a registered office in Beijing, Shanghai, Guangzhou, or Shenzhen is considered a nationwide business and is assigned a value of 1. An underwriter whose registered office is anywhere else is considered a regional business and is assigned a value of 0. A nationwide business has a valid reputation, while the reputation of a regional business is not significant.

2. Listing: Underwriters are divided into listed and nonlisted companies. A value of 1 is assigned to a listed company and 0 to nonlisted company.

3. Ranking: An underwriter in category A (AAA, AA or A) according to the CSRC's Classified Oversight Regulations for Securities Firms (revised in 2010) has a valid reputation and is assigned a value of 1. Any other category is assigned 0.

4. Securities underwriting volume: According to the Securities Association of China's (SAC) underwriting-volume-based annual underwriter rankings (2008 to 2010), the top 18 in each of the three years (about one-third of the total of 55 on annual average) are selected. By cross referencing, we cross out the same underwriter when it shows up for the second or third time, to reduce the final number to 28. We believe that the 28 underwriters have an advantage over the others in regard to goodwill, and therefore a value of 1 is assigned to them, and 0 to others.

5. Market share: We work out the subtotals of all lead underwriters for sample listed companies. We believe that an underwriter whose underwriting is above the average in regard to the number of IPO companies has a market share that may contribute to its goodwill and therefore has a reputation. A value of 1 is assigned to these underwriters and 0 is assigned to the others.

All the goodwill variables are dummy variables. An underwriter that is assigned a value of 1 is considered to have a reputation and the capabilities to make an effective valuation of an IPO company. The IPO premium rate on the securities of an underwriter should be lower than the premium rates on other securities. For the purpose of explaining the IPO premium rate, we also select the following four control variables:

1. Previous year's return on equity (ROE)

2. Size of public offering

3. Number of years between incorporation and IPO

4. First day's logarithmic return (calculated the same way as the IPO premium rate)

In addition to these four variables, we also add to the control variables the two dummies: "year" and "industry" (as per the CSRC's industrial classification standards). See Table 2.4 for details.

Sample and Data We split the A-share market into the main board, small and medium-sized enterprise (SME) board, and GEM and then create three subsamples for respective regression testing in the following two periods: one period between October 30, 2009 (the official launch of the GEM) and November 25, 2011, for the GEM sample; and one period between

TABLE 2.4 Table of Independent Variables

Table of Independent Variables

TABLE 2.5 Results of Hypothesis Test for the Mean in the GEM

District

Public

Level

Amount

MS

Goodwill variable = 0

0.305

0.301

0.358

0.355

0.330

Goodwill variable = 1

0.286

0.270

0.268

0.265

0.274

lvalue (T<,t) (2-tailed)

0.591

0.375

0.015**

0.011**

0.105

Note: * * *, * *, and * denote significance at the respective confidence level of 1 percent, 5 percent, and 10 percent.

June 19, 2006 (the restart of IPOs) and November 25, 2011, for the main board and SME board. The descriptive statistics of the three subsamples mentioned previously are as follows: For the GEM, the sample size is 259 (excluding those without available data for the variables, similarly from here on). A hypothesis test was done on sample data for the mean IPO premium rates of different variables. The results are shown in Table 2.5.

We can see at this point that considering all the goodwill variables, the mean IPO premium rates on securities underwritten by reputable underwriters are all lower than the rates on securities underwritten by underwriters that are considered to have no reputation. The p-values of the ranking and underwriting volume variables are significant at the 5 percent confidence level, which indicates that there is a significant difference between the mean values of 0/1 variables. The p-value of the market share variable is also significant at the confidence level slightly above 10 percent. We can at this point conclude that the goodwill mechanism works for reputable underwriters in the GEM.

For the Small and Medium-Sized Enterprise (SME) Board, the sample size is 523. The results of the hypothesis test for the mean are shown in Table 2.6.

We can see at this point that there is no real difference between the mean of goodwill variables for the IPO premium rates on securities underwritten by underwriters who have a reputation and those who do not. The mean of all variables also failed in the T-tests for variance. We may therefore

TABLE 2.6 Results of Hypothesis Test for the Mean in the SME Board Market

District

Public

Level

Amount

MS

Goodwill variable = 0

0.553

0.543

0.543

0.525

0.535

Goodwill variable = 1

0.538

0.541

0.542

0.548

0.544

iJ-value (T<t) (2-tailed)

0.706

0.951

0.966

0.579

0.828

TABLE 2.7 Results of Hypothesis Test for the Mean in the Main Board Market

District

Public

Level

Amount

MS

Goodwill variable = 0

0.247

0.309

0.301

0.174

0.313

Goodwill variable = 1

0.361

0.454

0.353

0.367

0.361

iJ-value (T<,t) (2-tailed)

0.258

0.063*

0.588

0.075

0.515

conclude that the goodwill mechanism does not work in the SME board market.

For the Main Board, the sample size is 90. The results of the hypothesis test for the mean are shown in Table 2.7.

We can see at this point that the mean IPO premium rates on securities underwritten by underwriters that have a reputation is higher, rather than lower, than such rates on securities underwritten by underwriters that do not have a reputation. In particular, the mean of the listing and underwriting volume variables are significant at the 10 percent confidence level in the T-tests for variance. Therefore, we preliminarily conclude that the goodwill mechanism increases, rather than lowers, the IPO premium rates in the main board market.

 
< Prev   CONTENTS   Next >